Journnl of Personality and Social Psychology 1985, Vol. 49, No. 1,250-263 .

Structure of Human Values: Testing the Adequacy of the Rokeach Value Survey

V. A. Braithwaite H. G. Law

_ Copyright 1985 by the American Psychological Association, Inc. 0022-3514/85/$00.75

Jllurrlll! of PcTSOMlity and Socilll Psycbology 1985. Vol. 49. No. 1,250-263

Copyright 1985 by the American Psychological Association, Inc. 0022-3514/85/$00.75

Structure of Human Values: Testing the Adequacy of the Rokeach Value Survey

V. A. Braithwaite Social Psychiatry Research Unit

The Australian National University

H. G. Law Department of Psychology

Universify of Queensland, Australia

Rokeach's (1973) Value Survey has received widespread use in the past decade, but little attempt has been made to examine the extent to which the 36 items provide comprehensive and representative coverage of the value domain. Our data provide qualified support for the comprehensiveness of the instrument. The major weaknesses in sampling involve the facets of physical well-being and individual rights. Other areas not represented are thriftiness and carefreeness. The need for multi-item indexes for value constructs are discussed, as are the advantages of a rating procedure over a ranking procedure from both psychometric and empirically valid perspectives. An alternative instrument based on the work of Rokeach is proposed.

In recent years, one of the most widely used instruments for measuring personal and social values has been the Rokeach Value Survey. Part of its popularity is undoubtedly due to the success researchers have had in finding specific values that differentiate var­ious political, religious, economic, genera­tional, and cultural groups and that relate to a range of social attitudes (Feather, 1975; Rokeach, 1973, 1979). The versatility of the instrument adds to its attractiveness. Feather (1972), for instance, used the survey to assess the match between own and perceived value systems, a variable that he related to personal adjustment.

A further advantage of the instrument is that success can be achieved relatively eco­nomically, because respondents need only deal with 36 concepts in all, each being conveyed by two or three short phrases. The task of rank ordering 18 end states of existence (terminal values) followed by 18 modes of conduct (instrumental values) in terms of their importance as guiding principles in life

Our thanks to Richard Bosly-Craft and Debra Rick­wood for assistance with data analysis and to Paul Duncan-Jones for his very helpful comments on the manuscript.

Requests for reprints should be sent to V. A. Braith­waite, National Health and .Medical Research Council, Social Psychiatry Research Unit, The Australian National University, Canberra, Australia 2600.

is, from all accounts, One that subjects com­plete quickly and easily.

In addition to these attractions, the Ro­keach Value Survey is one of the few instru­ments based on a well-articulated conceptu­alization of value. Like Kluckhohn (1951), Williams (1968), and Smith (1963), Rokeach (1973) located values in the realm of concep­tions of the desirable. Having done so, he followed Scott's (1965) example, elaborating the notion to provide clearer guidelines for the operationalizatioll of the construct. Scot!'s value attributes of absoluteness (applicability in all circumstances) and universality (accep­tance by others) were endorsed by Rokeach (1973) as he defined values as constructs that transcend specific situations and that are personally and socially preferable. Rokeach differed from Scott, however, in using value to incorporate not only modes of conduct, but also goals in life, the proviso being that such goals are not object specific. The goals are described as terminal values, thereby acquiring the quality of ultimacy referred to by Scott.

Hereupon follows Rokeach's (1973) major innovation. He set out a model of the belief system in which beliefs, attitudes, and values are clearly differentiated. The value construct is restricted to that special class of enduring beliefs concerning modes of conduct and end states of existence that transcend specific objects and situations and. that are personally

250

1 Ii' i

STRUCTURE· OF HUMAN VALUES 251

and socially preferable to an opposite mode of conduct or end state of existence. For the first time, the value domain has been clearly differentiated from that of other beliefs and attitudes.

The clear and restricted sampling frame for the Rokeach Value Survey carries with it the important implication that one can readily assess the adequacy of Rokeach's (1973) op­erationalization of the construct. Yet in spite of the widespread popularity of the instru­ment, empirical research addressing this issue is markedly absent. At the same time, re­searchers have been quick to express concern about Rokeach's operationalization. In par­ticular, the criteria for item selection have been criticized on the grounds of arbitrariness and subjectivity (Jones, Sensenig, & Ashmore, 1978; Keats & Keats, 1974; Kitwood & Smithers, 1975; Lynn, 1974). Indeed, Ro­keach (1973) himself acknowledged the overall procedure for selecting the 36 items to be "an intuitive one" (p. 30).

Rokeach (1973) chose the terminal values from several hundred that he compiled from the value literature, from personal experience, from the terminal values expressed by a representative sample of 100 inhabitants of an American city, and from those expressed by a small sample of graduate students. Items were then eliminated if they did not conform to Rokeach's definition of value or if they were regarded as semantically or empirically overlapping with others.

The point of departure for selecting the instrumental values was a list of 555 person­ality trait words that Anderson (1968) derived from the 18,000 Jrait names compiled by Allport and Odbert (1936). As with the ter­minal values, Rokeach's (1973) final selection was based on a long list of criteria. The most important criteria involved eliminating se­mantically or empirically equivalent items; selecting those considered to be important across culture, status, and sex; and avoiding values that would be strongly linked with a social desirability response bias (Rokeach, 1971, pp. 23-24). .

Directly addressing the issue of the com­prehensiveness and representativeness of the Rokeach Value Survey is Jones et al.'s(1978) comparison of spontaneously mentioned val­ues with Rokeach's (1973) 36 items. Jones et

al. concluded that the correspondence was poor, but it is of note that they restricted the spontaneously elicited values to the 36 that were most frequently mentioned, which con­stituted only 42% of those available for anal­ysis. Consequently, it would be fallacious to assume that Jones et al.'s items constitute a representative sampling of the value domain against which Rokeach's values should be compared.

In defending his instrument, Rokeach (1973) maintained that the final 36 items provide a "reasonably comprehensive" cov­erage of the most important human values (p. 27) and that the values are "negligibly correlated with one anotlier" (p. 43). Gorsuch (1970), although expressing some reservations about the sampling of items, concurred that the values selected indeed "cover a broad spectrum" (p. 139) and, when compared with the empirically derived responses reported by Scott (1959), give an impression of represen­tativeness. Gorsuch noted, however, that the self-ipsatizing nature of the instrument would make the correlations between items less strong than they would be with other mea­surement techniques. Thus the relative inde­pendence of the items and the absence of any strong underlying structure noted by both Rokeach (1973) and Feather and Peay (1975) may be more a function of the rank ordering task than of the values being ranked.

The rank ordering task has met with crit­icism on other grounds. Apart from the awk­wardness associated with the analysis of ip­sative data, researchers have questioned the meaningfulness of the task (Gorsuch, 1970; Keats & Keats, 1974; Kitwood & Smithers, 1975; Lynn, 1974). How does the researcher know from the data if the respondent endorses the total set of values or focuses on one or two at the expense of others? If values are not hierarchically organized, or if several values occupy the same level in the hierarchy, how does the individual respond to the task? If value constructs are being measured through single items, do differences between individuals Hreflect variations in linguistic usage rather than variations in underlying constructs" (Gorsuch, 1970, p. 139)?

Certainly the success with which the Ro­keach Value Survey has been used vindicates the item sampling procedures and the mea-

252 V. A. BRAITHWAITE AND H. G. LAW

surement technique to some extent. None­theless, the question of the comprehensiveness and representativeness of the items and the suitability of the hierarchical model are em­pirical questions that need to be subjected to more systematic investigation.

Research Goals

The starting point for our analysis of the adequacy of Rokeach's item set was the cri­teria for "comprehensiveness and represen­tativeness." Following McKennell (1974), we defined comprehensiveness and representa­tiveness in terms of the values that are salient for a particular population (in this case, inhabitants of a large Australian city). Such values can only be known to the researcher through consultation with a representative sample of informants, Reliance on literature searches; on previous questionnaires, or on the researcher's intuitions does not necessarily result in the identification of values that are meaningfully used by the population of in­terest. Similarly, the adequacy of the hierar­chical model must be evaluated ultimately by its concordance with the value priority struc­ture of individuals in the general population.

Thus our first task was to elicit from a sample of respondents the goals in life and ways of behaving that serve as guiding prin­ciples in their lives, and to understand how they are organized-that is, the priorities assigned to some over others. Like Rokeach (1973), we made no attempt to differentiate goals as means from goals as ends at the operational leveL Thus in this. research, as in that of Rokeach, the label terminal value is a misnomer. "Ways of behaving" refer to any Characteristic that describes an individual's manner of conduct in either a social or nonsocial setting and follows the conceptu­alization first outlined by Lovejoy (1950) and later adopted by Rokeach. At a conceptual level, we also acknowledged that a way of behaving may become a goal for a particular individuaL Not all goals, however, constitute ways of behaving. These guidelines provided the basis for distinguishing goals in life and ways of behaving. .

With a more comprehensive and represen­tative item set and some knowledge of priority setting in hand, one can achieve four further

aims. First, one can examine interrelations among the items to identify the major di­mensions underlying the value domain, It is reasonable to assume that values do not exist in isolation and that many of the items in a new instrument are interrelated, Second, by locating Rokeach's (1973) values within this new framework, one can on an empirical basis evaluate criticisms relating to the com­prehensiveness and representativeness of items in the Rokeach Value Survey. Third, this research paradigm provides an opportunity to investigate alternatives to single-item mea­sures. Rokeach's (1973) approach deviated from the well-established psychometric prin­ciple. of relying on several different measures when operationalizing a psychological con­struct. The inherent dangers in such a practice cannot be readily dismissed when the objects of the measurement exercise, values, are characterized by such a high degree of ab­straction. Such abstraction is bound to gen­erate problems of item ambiguity, leaving both the researcher and respondent with flex­ibility in interpretation. If researchers find that several measures correspond to Rokeach's single-item measures, they will have made' progress, toward more clearly defining the nature of the psychological constructs repre­sented in the Rokeach Value Survey.

Fourth, data on the major dimensions of human valuing and on priority setting should provide the basis for a new value instrument, incorporating Rokeach's (1973) principal concepts, but at the same time extending and amplifying them.

Experiment I: Development of Three Inventories

Method

Subjects. A sample of 115 adults was selected from the electoral roll for a single electoral division in the city of Brisbane, Australia. The division was chosen because it was characterized by a demographically heterogeneous population living in a relatively confined geographical area. Given compulsory electoral enrollment for all over the age of 18, this procedure provided a satisfactory way of obtaining a sample stratified on sex and occupation.

Of the 115 adults chosen, 73 (63%) participated in the study, 27 (24%) refused to participate, and 15 (13%) no longer resided at the given address. This sample constituted a wide cross-section of the community. The sample was 48% male. Twenty-seven percent of the sample were less than 30 years of age, 37·% were between 30 and 49 years

'I

STRUCTURE OF HUMAN VALUES

of age, and 36% were 50 or over. Of the 66% who were in paid employment, 17% had professional or managerial occupations, 26% clerical or sales, 19% trades, and 38% semiskilled or unskilled work.

Procedure. Intensive semistructured interviews were conducted with each subject. An interviews shared the same starting point: the Rokeach Value Survey. This approach offered a number of advantages. First, it was considerably easier to denne the universe of content through example than in abstract terms. Second, it provided valuable data on subjects' reactions to the Rokeach Value Survey. Interviewees were asked to com­ment on the nature of the task, the clarity of -items, and the instrument's comprehensiveness. Third, the survey eased respondents into a discussion of their own values and their value priorities, a topic that many broached with some hesitation. Initial reticence seemed most often due to subjects not being experienced in verbalizing their thoughts on such matters. Values were very much takenw forwgranted phenomena.

In developing a new instrument, there is always the danger that the sUbjective judgments of the researcher distort the way in which interviewees define the content domain. In order to reduce this effect, successive drafts of the instrument were returned to 6 of the suQiects who volunteered to act as informants and critics. In addition, 6 university students acted in the same capacity because the instrument was intended for use in this population as well.

Results

On the basis of the interviews, four modi­fications were made fo the Rokeach Value Survey. First, additional values emerged from discussion with participants, and many of Rokeach's (l973) original items were broken into components, which made them narrower in scope. This avoided most of the ambiguities perceived by respondents in the original items. Second, a rating scale became the preferred mode for responding. Apart from being a change of necessity with the increase in the number of values in the scale, the interview data suggested that it was also a desirable change. Although all respondents were able to produce a rank ordering of the items, some values were considered equally impor­tant, some were not to -be compared, and priority was sometimes determined by the situation.

Furthermore, most respondents (83%) per­ceived some of Rokeach's (I 973} values to be interrelated. Two strategies emerged for deal­i ng with this problem and for producing a single hierarchy of values. The approach used by 54% was to group the related values together in the hierarchy. Criteria for deciding

priority included not only importance, but also which was the most general or which was the means to achieving others. The second strategy, used by 20%, was to optimize the attainment of several values by placing one representative of each value group at the top of the hierarchy while relegating the remain­der to the bottom. In other words, interrela­tions perceived among values influenced the importance assigned to a particular value. The remaining 9% who saw value relation­ships used a mixture of the two strategies.

Although the possible effects of respon­dents' ranking strategies on correlational structure have not yet been investigated em­pirically, they do highlight the advantages of a rating procedure. After discussions with respondents, we adopted an asymmetrical 7-point rating scale:

1. I reject this as a guiding principle in my life. . 2. I am inclined to reject this as a guiding-principle in

my life. 3. I neither reject nor accept this as a guiding principle

in my life. 4. I am inclined to accept this as a guiding principle in

my life. 5. I accept this as important as a guiding principle. in

my life. 6. I accept this as very important as a guiding principle

in my life. 7. I accept this as of the greatest importance as a guiding

principle in my life.

The asymmetrical scale, first suggested as an option by Gorsuch (l970), involves finer dis~ criminations by respondents on the positive end, because distributions tended to be neg­atively skewed. The fact that respondents found most of the values highly desirable was not surprising, given that values are widely accepted as phenomena transmitted by soci­ety's major institutions (Kluckhohn, 1951; Rokeach, 1973).

The third deviation from Rokeach's (1973) instrument was that the goals in life were presented in two parts rather than one. In­terviewees consistently differentiated personal goals (e.g., a sense of accomplishment) from societal goals (e.g., a world of peace), regard­ing the latter as something they did not have direct influence over. This led to confusi()n as to whether they should be ranked according to the respondents' beliefs or their actions to achieve the goal. As a result, the 18 societal goals were separated from the 36 personal I

I ~

I I

254 V. A. BRAITHWAITE AND H. G. LAW

goals and given a new set of instructions. We ensured greater uniformity in interpretation in the Social Values Inventory by asking subjects to judge the importance of the soci­etal goals in guiding not only their actions, but also their judgments about national or international events.

Fourth, for the 71-item Mode Values In­ventory, the measure of Rokeach's (1973) instrumental value system, the word being was inserted in front of each way of behaving. This served to remind subjects to evaluate the items as. behavior patterns that they may or may not try to emulate, rather than as traits that describe the sort of person they are.

The test-retest reliabilities for the items were examined over a 4-week period with a sample of 208 university students. For the Goal Values Inventory, the coefficients ranged from .43 to .94; the median was .62. The range for the Mode Values Inventory was .43 to .90 (Mdn = .61), whereas that for the Social Values Inventory was.46 to .92 (Mdn =

.62). These reliability coefficients are com' parable to those reported by Rokeach (1973) and Feather (1971) for the Rokeach Value Survey.

Experiment 2: Factor Structures of the Three Inventories

Method Subjects. We -used three independent- samples to

investigate factor structure:' one from the general popu­lation and two from the student population. The General Population Study was based on a stratified random sample of 483 adult inhabitants of Brisbane, Australia. Details of the procedure and nature of the sample have been provided elsewhere (Braithwaite, 1982). The students who participated in the other studies were introductory psychology students at the University of Queensland, Australia. They numbered 208 and 480 in Student Studies 1 and 2, respectively.

Procedure. In the General Population Study, ques­tionnaires were mailed to respondents on the understand­ing that a research worker would call 2 weeks later to collect the completed questionnaire and to answer any queries. As well as responding to the Goal, Mode, and Social Values Inventories, participants were asked to provide basic sociodemographic information. Of those contacted, 61% participated.

For the student studies, questionnaires were adminis­tered in a classroom situation. In the Student Study 1, only the Goal, Mode, and Social Values Inventories were completed. In the Student Study 2, the value inventories were part of a battery of tests administered over a 3-

week period. Included in the battery was a rating form of the Rokeach -Value Survey. Because a major goal of the research was to examine the adequacy of the items in the Rokeach Value Survey, the response format used in the value instruments was identical. The order of

. presentation of the tests in the battery was systematically varied.

In all three studies, the value inventories were under­taken in a fixed order: first the Goal Values Inventory, then the Mode Values Inventory, and finally the Social Values Inventory. All respondents were assured of ano­nymity.

Results

To identify the dimensions underlying the more comprehensive and representative value sample derived in this research, we factor analyzed the three data sets separately, giving serious attention to only those factors that showed some stability across studies. We in­tercorrelated items by using the Pearson product-moment correlation coefficient, sub­jected the resulting matrix to an alpha factor analysis, and subsequently rotated the solution by using the promax procedure (k = 4). The resulting solution was comparable to those obtained with other factoring and rotational procedures, 1 but produced a better simple structure. The number of factors criterion used in each study was an eigenvalue cutoff of I, Guttman's lower bound for the number of factors (Guttman, 1954; Kaiser, 1960). The factor structure showed most stability in spite of changing the number of factors for rotation in the vicinity of this cutoff.

Comparisons across data sets were made by means of the coefficient of congruence (Burt, 1948; Tucker, 1951; Wrigley & Neu­haus, 1955) and the salient variable similarity index (Cattell, 1949; Cattell & Baggaley, 1960). The coefficient of congruence indicates the degree of relation between loadings on pairs of supposedly matching factors. The salient variable similarity index is a nonpara­metric technique that provides the probability that identical variables defining two factors of interest have occurred by chance. Given the exploratory nature of the research, we considered comparisons of factors on an in­dividual basis preferable to comparisons of

I Other procedures used were principal axes factor analysis and Guttman-Lingoes nonmetric factor analysis (Lingoes &,Guttman"1967) and varimax rotation.

I

STRUCTURE OF HUMAN VALUES 255

Table I Coefficients oj Congruence (C) and Salient Variable Similarity (S) Indexes for Factors Stable Across the General Population (GP) and Student (Sl, S2) Studies

Comparison

GP-Sl GP-S2 SI-S2

Factor C S C S C S

Goal and Social Values Inventories International harmony and equality .93 6/7 .96 6/7 .94 517 National strength and order .89 6/7 .90 517 .93 717 Traditional religiosity .86 517 .89 4/7 .87 617 Personal growth and and inner

harmony .78 517 .91 .617 .85 517 Physical well-being .82 317' .88 4/7 .91 4/7

Mode Values Inventory Positive orientation to others .91 7/10 .92 6/10 .86 7/10 Competence and effectiveness .89 6/10 .93 8/10 .88 7/10 Propriety in dress and manners .86 7/10 .92 8/10 .85 6/10 Religious commitment .79 5/10 .85 8/10 .68 6/10 Assertiveness .73 4/10 .83 6/10 .77 4/10 Withdrawal from others .78 7/10 .82 6/10 .73 7/10

Note. S based on the number of coincidences found between the most saliently loading variables on pairs of factors­(marker variables excluded) and on the associated probability values given by Cattell and Baggaley (1960). The seven highest loadings were considered for the Goal and Social Values Inventories; the 10 highest loadings were considered for the Mode Values Inventory. * p = .01. For all other,S values, p < .005.

factor solutions achieved through procrustes rotation methods or confirmatory factor analysis.

Within each data set, the Goal and Social Values Inventories were analyzed together; the Mode Values Inventory was analyzed: separately. Although separate instructions were given for the goal and social, values, they were analyzed together to maximize compatibility with Rokeach's (I973) concep­tualization of two value systems: a terminal value system and an Instrumental value system.

Goal and Social Values Inventories. Be­tween 46% and 53% of the total variance in the item set was accounted for in the three factor solutions. Nine factors were regarded as representing potentially important value dimensions, tbough indexes of stability were consistently satisfactory for only five of them (sec Table I). The factors, whose items and loadings are given in Table 2, were interpreted us representing (a) international harmony and equality (representing a political ideology directed toward achieving a more cooperative,

equitable, and humanistic social order); (b) national strength and order (emphasizing the attainment of economic and political might together with internal order); (c) traditional religiosity; (d) personal growth and inner harmony; (e) physical well-being; (f) secure and satisfying interpersonal relationships; (g) social standing; (h) social stimulation; and (i) individual rights. For the last four factors, coefficients of congruence were not sufficiently high for us to claim a match, though the salient variable similarity indexes were sig­nificant at the .01 level for all except social stimulation.

Mode Values Inventory. Between 49% and 54% of the total variance was accounted for by the three factor analyses of the Mode Values Inventory. We regarded ten dimensions as potentially important because of their re­emergence across the three data sets. Once again, however, not all satisfied both criteria for stability across studies. (See Table I for

/ those with satisfactory coefficients of congru­ence and salient variable similarity indexes.) The factors were labeled as follows (their

256 V. A. BRAITHWAITE AND H. G. LAW

Comparison of the Three Inventories With the'Rokeach Value Survey

items and loadings are shown in Table 3): (a) positive orientation toward others (describing ways of interacting with others that reflect warmth, concern, and kindness); (b) compe­tence and effectiveness (concerned with the capacity to get a job done; items incorporate what are essentially ability items with a desire to perform a task well); (c) propriety in dress and manners (behaving in a conventionally upright and decent manner); (d) religious commitment (forsaking self-interest and pur­suing a higher cause); (e) assertiveness; (f) withdrawal from others; (g) carefreeness; (h) honesty; (i) thriftiness; and U) getting ahead. For the last four factors, neither stability coefficient proved consistently satisfactory over the three data sets.

Using data from the Student Study 2 (n = 480), we formed scales from the items listed in Tables 2 and 3 as defining the most stable factors. Fifteen showed sufficient internal consistency to be considered measures of specific value constructs. From the items that were not sufficiently intercorrelated to form a scale, single items were selected as the best available representatives of underlying con­structs. The scales with their alpha reliability coefficients and the items with their test­retest reliabilities appear in Tables 4 and 5.

The scales and single-item measures were subsequently related to Rokeach's (1973) 36

Table 2 Factors, Items, and Loadings for the Goal and Social Values Inventories

Item

International Harmony and Equality A good life for others: improving the welfare of all people in need Rule by the people: involvement by all citizens in making decisions that affect their

community International cooperation: having all nations working together to help each other Social progress and social reform: readiness to change our way of life for the better A world at peace: being free from war and conflict A world of beauty: having the beauty of nature and of the arts (music, literature, art, etc.) Human dignity: allowing each individual to be treafed as someone of worth Equal opportunity for all: giving everyone an equal chance in life Greater economic equality: lessening the gap between the rich and the poor Preserving the natural environment: preventing the destruction of nature's beauty and

resources

National Strength and Order National greatness: being a united, strong, independent, and powerful nation National economic development: having greater economic progress and prosperity for the

nation . The rule of law: punishing the guilty and protecting the innocent National security: protection of your nation from enemies

Traditional religiosity Salvation: being saved from your sins and at peace with God Religious or mystical experience: being at one with God or the universe Upholding traditional sexual moral standards: opposing sexual permissiveness and

pornography Sexual intimacy: having a satisfying sexual relationship

Personal growth and inner harmony Self-knowledge or self-insight: being more aware of what sort of person you are The pursuit of knowledge: always trying to find out new things about the world we live in Inner harmony: feeling free of conflict within yourself Self-improvement: striving to be a better person Wisdom: having a mature understanding .of life Self-respect: believing in your own worth

Loadings

.59, .60, .72

.53, .40, .56

.77, .70, .66

.68, .57, .61

.53, .73, .55

.38, .64, .34 .

.33, .35, .47

.53, .63, .64

.59, .65, .64

.34, .61, .38

.74, .68, .68

,67, .63, .70 .58, .59, .58 .70, .86, .80

.85, .86, .87

.80, .66, .73

.53, .62, .53 -.19, -.39, -.43

.56, .66, .66

.37, .65, .48

.46, .48, .51

.36, .44, .42

.44, .27, .46

.44, .13, .49

1 i'

t I ~'

t !

I

STRUCTURE OF HUMAN VALUES 257

Table 2 (continued)

Physical well-being . Physical development: being physically fit Good health: physical well-being

Item

Physical exercise: taking part in energetic activity

Secure and satisfying interpersonal relationships

Loadings

.64, .83, .82

.46, .67, .62

.56, .76, .69

Mature love: having a relationship of deep and lasting affection True friendship; having genuine and close friends

.70,

.41, .74, .47 .54, .59

Personal support: knowing that there is someone to take care of you Security for loved ones: taking care of loved ones

.38,

.36, .63, .37. .47, .36

Acceptance by others: feeling that yOll belong .53, .18, .59

Social Standing Recognition. by the community: having high standing in the community Economic prosperity: being financially well off

.49,

.37, .56, .52 .21, .39

Authority: having power to influence others and control decisions .13, .42, .39

Social Stimulation An active social life: mixing with other people .51, .62, .56 An exciting life: a life full of new experiences or adventures .61, .31, .54

Individual Rights Privacy for yourself: being able to keep your business to yourself .45, .65, .28 A sense of ownership: knowing that the things you need and use belong to you A leisurely life: being free from pressure and stressa

AI; .54,

.45, .30

.09, .03 Carefree enjoyment: being free to indulge in the pleasures of life .42, .07, .03 The protection of human life: taking care to preserve your own life and the life of othersb

Comfort but not luxury: being satisfied with the simple pleasures of lifeb .00, .18,

.36, .40

.35, .33

Note. Factor pattern loadings from the General Population Study, the Student Study, and the Student 2 Study, respectively. Complete factor solutions are available from the authors on request. a Loadings are salient for the general population, suggesting an interpretation of this factor as individual rights and the good life. b Loadings are salient for the student samples, suggesting an interpretation of this factor as individual rights and basic necessities.

terminal and instrumental values. From the two instruments, items that were almost identical or that could be considered seman­tically equivalent were used as marker vari­ables in order to hypothesize matches between the factor analytically derived value constructs and those measured by the Rokeach Value Survey.

Terminal Values

Pearson product-moment correlation coef­ficients were calculated between the newly derived value constructs and their hypothe­sized counterparts in Rokeach's (1973) in­strument (see Table 4). Seven of the nine factors are represented in the Rokeach Value Survey, and four of these-International Har­mony and Equality, Personal Growth and

Inner Harmony, Secure and Satisfying Inter­personal Relationships, and Social Standing­are represented by more than one item. Those with only single-item representation are Na­tional Strength and Order, Traditional Reli­giosity, and Social Stimulation.

Factors with no clear counterparts at all were Physical Well-Being and Individual Rights and Basic Necessities (see Footnote b, Table 2). Physical Well-Being was not in­cluded in any guise in the Rokeach Value ·Survey. The item in the instrument with which it correlated most highly was Social Recognition (respect, admiration), r(456) = .30, p < .001. Individual Rights and Basic Necessities, on the other hand, might be expected to bear some relation to the Rokeach Value Survey item Freedom (independence, free choice), the common ground being in-

258 V. A. BRAITHWAITE AND H. G. LAW

dividualism. Representation through this item, however, is not impressive, r( 456) ~ . 29, p < .001.

Regarding the items of the Value Survey, three terminal values could not be successfully linked a priori with the newly derived factors. The first, Freedom, has already been men­tioned as a possible correlate of Individual Rights and Basic Necessities. It was, however, far more strongly related to International Harmony and Equality, r( 456) ~ .45, p < .001. Rokeach's (1973) measure of freedom appears to connote democratization rather than individualism (Braithwaite, 1982). Given that Rokeach conceptualized freedom as rep-

Table 3

resenting a value dimension that was orthog­onal to equality,. the construct validity of this item can be justifiably questioned .

The second of Rokeach's (1973) values that is not represented in the newly derived factor structure is Pleasure (an enjoyable, leisurely life). Pleasure was represented in the Goal Values Inventory by two items: A Lei­surely Life (being free from pressure and stress) and Carefree Enjoyment (being free to indulgein the pleasures oflife). They played a central role in defining the Individual Rights factor in the General Population Study, but defined a separate specific factor in the student data sets. Specificity, however, is not synon-

Factors, Items, and Loadings for the .Mode Values. Inventory

Item

A Positive Orientation to Others Tolerant: accepting others even though they may be different from you Helpful: always ready to assist others Forgiving: willing to pardon others Giving others a fair go: giving others a chance Tactful: being able to deal with touchy situations without offending others Considerate: being thoughtful of other people's feelings Cooperative: being able to work in harmony with others Loving: showing genuine affection Trusting: having faith in others Grateful: being appreciative Understanding: able to share another's feelings Friendly: being neighborly Generous: sharing what you have with others

Competence and Effectiveness Bright: being quick thinking AQ.aptable: adjusting to change easily Competent: being capable Resourceful: being clever at finding ways to achieve a goal Self-disciplined: being self.controlled Efficient: always using the best method t() get the best results Realistic: seeing each situation as it really is Knowledgeable: being well informed Perservering: not giving up in spite of difficulties Progressive: being prepared to accept and support new things Conscientious: being hardworking Logical: being rational . Showing foresight: thi:nking and seeing ahead

Propriety in Dress and Manners Polite: being 'well-mannered Patriotic: being loyal to your country. Prompt: being on time Refined: never being coarse or vulgar Clean: not having dirty habits Neat: being tidy Reliable: being dependable

Loadings

.48, .65, .58

.58, .50, .56

.54, .62, .68

.46, .61, .53

.47, .56, .34

.61, .67, .63

.38, .37, .42

.39, .47, .47

.37, .40, .30

.48, .31, .52

.54, .43, .66

.52, .39, .43

.52, .53, .67

.45, .38, .48

.49, .32, .48

.60, .54, .54

.57, .70, .70

.37, .48, .34

.54, .61, .72

.64, .34, .61

.54, .52, .54

.42, .57, .33

.43, .44, .37

.33, .54, .32

.52, .47, .56

.49, .76, .45

.53, .65, .55

.44, .41, .40

.51, .24, .56

.60, .64, .54

.68, .63, .57

.67, .73, .63

.26, .19, .44

I I I i

I I

i

STRUCTURE OF HUMAN VALUES 259

Table 3 (continued)

Item Loadings

Religious, Commitment Committed: being dedicated to a cause .57, .67, .50 Devout: following your religious faith conscientiously Self-sacrificing: putting the interest of others before your own

.62,

.48, .51, .50 .57, .48

Idealistic: living according to how things should be rather than how things are .42, .03, .65

Assertiveness Standing'up for YOUf beliefs: defending your beliefs no matter who opposes them Having your say: confidently stating your q:,inions

.58,

.64, .42, .43 .60, .50

Determined: standing by your decisions firmly .55, .70, .68

Withdrawal From Others Keeping to yourself: being content with your own company Independent: doing things on your own

.57,

.39, .68, .52 .34, .45

Carefreeness Acting on impulse: doing things on the spur of the moment Spontaneous: doing what comes naturally

.42,

.32, .64, .67 .55, .53

Cautious: not rushing into things

Honesty Open: not hiding anything from~anyone Honest: never cheating or lying

Thriftiness Thrifty: being careful in spending money

-.05, -.36, -.34

.51, .48, .43

.26, .43, .40

.52, .57, .46 Never missing a chance: taking advantage of every opportunity that comes your way .13, .56, .21

Getting Ahead Ambitious: being eager to do well .41, .23, .63 Competitive: always trying to do better than others .61, .53, .61

Note. Factor patt~rn loadings from the General Population Study, the Student Study respectively. Complete fa,ctof solutions are available from the authors on request.

1, and the Student Study 2,

. ymous with unimportance. Until further re­search has been conducted to clarify the interrelations and status of the pleasure values, there is little justification for omitting them from instruments such as the Rokeach Value Survey.

The third terminal value that could not be placed within the newly <lerived factor struc­ture was Happiness (contentedness). The~cor­responding item in the Goal Values Inventory, Happiness (feeling pleased with the life you are leading), was found to have low to mod­erate loadings on a number of factors. Mir­roring this finding, Rokeach's (1973) item correlated between .2 and .3 with eight of the ten factor analytically derived scales. Loose yet interpretable associations with several value constructs support the view that hap­piness is a more fundamental value than the majority of items in the value instrument

(Braithwaite, Law, & Braithwaite, 1984). Be­cause of its fundamental nature, happiness deserves representation in any instrument with which one seeks to provide a means of assessing individual or cultural values.

Instrumental Values

As can be seen trom Table 5, seven of the ten constructs derived from the Mode Values Inventory have been represented in the Ro­keach Value Survey, and those that have been overlooked are not serious omissions.

Religious commitment is one of the few ways of behaving that is represented in the Goal Values Inventory. Predictably, Tradi­tional Religiosity and Religious Commitment correlated highly, r( 456) = .61, P < .00 I. The single-item measures, Acting on Impulse and Being Thrifty, had no counterparts in the

260 v. A. BRAITHWAITE AND H. G. LAW

Table 4 Correlations a/the Goal and Social Values Inventories Indexes With Corresponding Items From the Rokeach Value Survey

Goal and Social Values Inventories Index

International Harmony and Equality (a = .86)

National Strength and Order (0: = .66)

Traditional Religiosity (a = .75)

Personal Growth and Inner Harmony (a = .76)

Physical Well-Being (a = .79)

Secure and Satisfying Interpersonal Relationships (a ~ .70)

Social Standing (0: = .70)

Social Stimulation (a = .58)

Individual Rights and Basic Necessities (a = .55)

Rokeach Value Survey item

A world at peace: free of war and conflict A world of beauty: beauty of nature and the arts Equality: brotherhood, equal opportunity for all

National security: protection from attack

Salvation: saved, eternal life

A sense of accomplishment: lasting contribution Inner harmony: freedom from inner conflict SelfMrespect: selfMesteem Wisdom: a mature understanding of life

Family security: taking care of loved ones Mature love: sexual and spiritual intimacy True friendship: close companionship

A comfortable life: a prosperous life Social recognition: respect, admiration

An exciting life: a stimulating, active life

r

.50

.52

.61

.59

.79

.33

.27

.38·

.51

.42

.36

.48

.48

.56

.' .44

Note. N = 458. All r values are significant at the .001 level.

Rokeach Value Survey. These values appear to be unrelated to other major value clusters, and their status at this stage remains unclear.

With regard to multi-item representation, three scales-A Positive Orientation to Oth­ers, Competence and Effectiveness, and Pro­priety in Dress and Manners-are represented by sets of items in the Rokeach Value Survey. Two other scales-Assertiveness and Getting Ahead-are covered by single items. The remaining two constructs, Being Honest and Being Independent, have single-item repre­sentation in both the Rokeach instrument and the Mode Values Inventory.

Discussion

We conducted this study to answer two questions concerning the Rokeach Value Sur­vey: (a) Does the instrument provide a com­prehensive and representative coverage of the major value constructs? (b) In rank ordering the values, does one use the· optimal mea­surement technique? We initiaJly approached the second question by examining the. fit between Rokeach's (1973) hierarchical model

and self-reports of priority setting from a sample from the general population. Subse­quently, we investigated the option of mea­suring each value construct through several items rather than one.

In general, the Rokeach Value Survey is successful in covering the many and varied facets of the value domain. The major excep­. tion is the neglect of values relating to physical development and weJl-being. This is clearly an oversight in the development of the instru­ment, particularly given the representation of values associated with mental health (e.g., inner harmony and self-respect). Such values could be expected to have relevance to a number of substantive research areas, not the least important of which woilld be the social determinants of physical iJlness.

A second problem area relates to basic human rights such as dignity, privacy, the protection of human life, and freedom. None of these are adequately represented in the Rokeach Value Survey and yet are among the most fundamental tenets of our society.

The importance of other neglected values is more ambiguous and remC:l.ins a question

I i

STRUCTURE OF HUMAN VALUES

for future research. In this category are the mode values related to thriftiness and care­freeness.

The inclusion of two of Rokeach's terminal values-pleasure and happiness-was not supported empirically by the results of our research, but their representation in the in­strument was defended. The different patterns of interrelations of the pleasure variables in the student and general population samples were considered worthy of further investiga­tion. They may, for instance, be attributable to the varying connotations attached to plea­sure by different age groups. In the case of happiness, the value is regarded as a basic building block, a value in terms of which other values are justified (Braithwaite, Law, & Braithwaite, 1984). Not surprisingly, there-

Table 5

fore, it was moderately related to a number of different value constellations rather than being strongly related to one.

With regard to the question of multi-item measurement, our data suggest that Rokeach (1973) actually did have more than one mea­sure of each of several constructs in his· instrument. There remains, however, a sig­nificant number of single-item indexes for such constructs as Pleasure, Happiness, Tra­ditional Religiosity, Social Stimulation, Na­tional Strength and Order, Assertiveness, Withdrawal From Others, Honesty, and Get­ting Ahead. In some cases, the items repre­senting these facets of the value domain have proven themselves both reliable and valid as single indicators. Salvation, for instance, falls into this category. The usefulness of others,

Correlations a/the Mode Values Inventory Indexes With Corresponding Items From the Rokeach Value Survey

Mode Values Inventory Index

A Positive Orientation to Others (a = .89)

Competence and Effectiveness (a == .90)

Propriety in Dress and Manners (a == .83)

Religious commitment (a = .68)

Assertiveness (a = .69)

Being Independent: Doing things on your own (test-retest r = .65)8

Acting on Impulse: doing things on the spur of the ~oment (test-retest r = .69)11

Being Honest: never ·cheating or lying (test-retest r = .60)3

Being Thrifty: being careful in spending money (test:-re.test r = .64)a

Getting Ahead (a = .67)

Note. All r values are significant at the .00 I level. a Single items chosen to represent the factor.

Rokeach Value Survey Item

Forgiving: willing to pardon--others Broadminded: openRminded Helpful: working for the welfare of others Loving: affectionate, tender Cheerful: light hearted, joyful

Capable: competent, effective Intellectual: intelligent, reflective Logical: consistent," rational SelfRcontrolled: restrained, selfRdisciplined Imaginative: daring, creative

Clean: neat, tidy Obedient: dutiful, respectful Polite: courteous, wellRmannered Responsible: dependable, reliable

Courageous: standing up for your beliefs

Independent: selfRreiiant, self-sufficient

Honest: 'sincere, truthful

Ambitious: hardworking, aspiring

r

.55

.37

.52

.48

.40

.45

.54

.55

.49

.44

.59

.47

.63

.57

.52

.55

.48

.55

262 V. A. BRAITHWAITE AND H. G. LAW

however, is more questionable. For instance, Rokeach's measures of honesty, independence, and pleasure are not particularly stable over a period of weeks (Feather, 1975; Rokeach, 1973).

The factor necessitating the use of single­item measures in Rokeach's (1973) instru­ment is the rank ordering task itself. Yet these data fail to demonstrate that a single rank ordering of values reflects the priorities operating for members of the general popu­lation. Individuals do hold some values as equally important; some values never corne into conflict, and others just aren't compared. In the absence of strong evidence to support Rokeach's hierarchical model, a rating pro­cedure with multi-item measures for each value construct becomes an attractive alter­native.

The findings of our research provide the basis for the development of such an instru­ment. These data provide psychometrically satisfactory measures of 14 value constructs. The number of items in these scales, their test-retest reliabilities, and their alpha reli­ability coefficients for the aggregated student samples and the general population saniple, respectively, are presented in Table 6.

Seven additional constructs need to be assessed. A psychometrically satisfactory scale to measure individual rights has, unfortu­nately, proven elusive, though these data pro, vide a sound basis for further developmental work. Refinements to the items should pro­duce a scale with sufficient stability and coherence. For the remaining six values, sin­gle-item measures must suffice at this time. Rokeach's (1973) items-(a} Pleasure (an en­joyable, leisurely life), (b) Happiness (con­tentedness), (c) Independent (self-reliant, self­sufficient), and (d) Honest (sincere, truthful)­are probably the best available measures of these constructs. The Mode Values Inventory produced two additional single-item indexes: Acting on Impulse (doing things on the spur of the moment) and Being Thrifty (being careful in spending money).

This raises the question of whether it is possible within the confines of Rokeach's (l973) conceptualization of value to derive multiple-item measures of these constructs, or, for that matter, to increase the number of items in some of the indexes mentioned

Table 6 Fourteen Value Constructs for Aggregated Student Samples and General Population Sample

No. Test-retest Alpha Variable items reliability reliability

1. International harmony and equality to .73 .85, .83

2. National strength and order 4 .81 .81, .83

3. Personal growth and_ inner harmony 6 .70 .74, .73

4. Physical well~being 3 .71 .79, .74 5. Secure and satisfying

interpersonal relationships 5 .71 .70, .68

6. Social standing 3 .77 .70, .65 7. Social stimulation 2 .58 .53, .53 8. Traditional religiosity 4 .93 .75, .70 9. Religious

commitment 4 .81 .66, .66 10. Positive orientation

toward others 13 .80 .89, .88 II. Competence and

effectiveness 13 .81 .89, .88 12. Propriety in dress

and manners 7 .84 .83, .82 13. Assertiveness 3 .68 .67, .72 14. Getting ahead 2 .72 .66, .59

earlier. Given that we undertook lengthy in­terviewing procedures to ensure adequate representation of the domain, deriving addi­tional measures of the construct may prove more difficult than might be assumed. The nuances of the English language may be such that there are only one or two options for expressing such values as thriftiness or plea­sure within Rokeach's framework. This does not, of course, preclude the possibility of inferring values from measures involving sets of ideal goals in life and ways of behaving that are more specific in their focus-state­ments that Rokeach would regard as attitu­dinal in nature.

In suggesting an alternative to the Rokeach Value Survey, we are not denigrating the usefulness of this instrument nor offering a panacea for the problems of value measure­ment. As outlined previously, the survey has clear conceptual advantages over earlier in­struments, it fared well in terms of its rep­resentativeness and comprehensiveness, and, unwittingly, it has even more than one- mea-

STRUCTURE OF HUMAN VALUES 263

sure of some value constructs. The self-ipsa­tizing nature of the instrument, however, is a feature that does not seem to be justified either psychometrically or in terms of empir­ical validity. This is not to say that the alternative procedure suggested here, rating the values, is without weaknesses. Indiscrim­inate use of the more favorable categories remains a problem, and the development of category labels and appropriate instructions to limit such behavior deserves high research p~iority. At the same time, overuse of positive categories is not at all surprising when one remembers what values are. With this in mind, one must guard against developing an instrument that forces discriminations for statistical neatness while failing to reflect psychological realities. After all, it may not be the holding of particular values but rather the ability to assign priorities among one's values that is the key to understanding the way in which values influence behavior.

References

Gorsuch, R. L. (1970). Rokeach's approach to value systems and. social compassion. Review oj Religious Research, Il, 139-143.

Guttman, L. (1954). Some necessary conditions for factor analysis. Psychometrika, 19, 149-161.

Jones, R. A., Sensenig, J., & Ashmore, R. D. (1978). Systems of values and their multidimensional repre­sentations. Multivariate Behavioural Research, 13. 255-270.

Kaiser, H. F. (1960). The application of electronic com­puters to factor analysis. Educational and Psychological Measurement, 20, 141-15l.

Keats, D. M., & Keats, J. A. (1974). Review of "The nature of human values" by M. Rokeach. Australian Journal of Psychology. 26, 164-165.

Kitwood, T. M., & Smithers, A. G. (1975). Measurement of human values: An appraisal of the work of Milton Rokeach. Educational Research, 17, 175-179.

Kluckhohn, C. K. M. (1951). Values and value orientation in the theory of.action. In T. Parsons & E. Shils (Eds.), Toward a general theory oj action (pp. 388-433). Cambridge, MA: Harvard University Press.

Lingoes, J. c., & Guttman, L. (1967). Non-metric factor analysis: A rank reducing alternative to linear factor. analysis. Mul!ivariate Behavioural Research, 2, 485-505.

Lovejoy, A. 0. (1950). Terminal and adjectival values. Journal of Philo,ophy. 47. 593-608.

Lynn, R. (1974). Review of "The nature of human values" by M. Rokeach. British Journal oj Psychology, 65, 453.

Allport, G. W., & Odbert, H. S. (1936). Tmit-namcs: A McKennell, A. C. (1974). Surveying attitude structures. psycho-lexical study. Psychological Monographs, 47 Amsterdam: Elsevier. (no. 211, p. 171). Rokeach, M. (1971). The measurement of values and

Anderson, N. H. (1968). Likableness ratings of 555 value systems. In G. Abcarian & J. W. Soule (Eds.), Personality-trait words. Journal oj Personality and Social psychology and political behavior. Columbus, Social Psychology, 9, 272-279. OR Charles Merrill.

}c'Braithwaite, V. A. (1982). The structure of soc~a.1 values: XRokca~h, M. (1973). The nature oj human values. New Validation of Rokeach's two value model. Br/flsh .Jollr- York: Free Press. nal oj Social Psychology, 21, 203-211. " Rokeach, M. (1979). Understanding human values .. New

Braithwaite, V. A., Law, H. G., & B.ra!thwultc, J. B. York: Free Press. (1984). Value systems: Perceived and 1I!/em..'d struclllre. X Scott, W. A. (1959). Empirical assessment of values and Unpublished manuscript. ideologies. American SOciological Review, 24, 299-

Burt, C. L. (1948). The factorial study of t~n.lpcrame.ntal 310. traits. British Journal oj Psychology (Slat/st/col S(!ctum), Scott, W. A. (1965). Values and organizations: A study 1, 178-203. .. ojj'raternities and sororities. Chicago: Rank McNally.

Cattell, R. B. (1949). A nole on factor mvananc~ ,md Smith M. 8. (1963). Personal values in the study of the identification of factors. British Journal oj Psy- Iive~. In R. W. White (Ed)., The study oj lives (pp. chology (Statistical Section), 2, 134-139. . 325-347). New York: Atherton.

Cattell, R. 8., & Ba?£aley, A. R. (1960). !hc sal.l~nt Tucker, L. R. (1951). A method· for synthesis of factor variable similarity mdex for factor match mg. IJrlllsh analysis studies. Personnel Research Section Report, Journal oJ Statistical Psychology. 13, 33-46, No 984 Washington DC: U,S, Department of the

Feather, N. T. (1971). Test-retest reli~bility ofindi~idual Ar~y.· , values and value systems. Australwn Psycholog/st, 6, Williams, R. M. J 1968). Values. In E. Sills (Ed.), Inter-181-188. . . . . national encyclopedia oJlhe social sciences (pp. 283-

Feather, N. T. (1972). Value slmdanty and school adJust- 287). New York: Macmillan. ment. Australian Journal oj ~sychologJ! 24, 193-2?8. Wrigley, C. S., & Neuhaus, J. (1955). The matching oj

Feather, N. T. (1975). Values In educatlOn and soclely. two sets oj Jactors (Contract Memorandum Report New York: Free Press. A-32). Urbana: University of Illinois.

Feather, N. T., & Peay, E. R. (1975). The structure of terminal and instrumental values: Dimensions and clusters. Australian Journal oj Psychology, 27, 151-164.

Received April 5, 1983 Revision received February- 28, 1984 •